1 free read left. Create a free account for unlimited reading + AI research.
Create free accountWhitepaper
DiGRA
Finite State Machines give predictable NPC control, while Agent-Based Models add dynamic, autonomous behavior. The hybrid aims to balance instructional consistency against real-world complexity.
This study introduces a novel hybrid model for adaptive scenario selection within serious games, specifically an educational RPG focused on software development. The core thesis posits that integrating Finite State Machines (FSMs) and Agent-Based Models (ABMs) can dynamically adjust Non-Player Character (NPC) behavior and game progression, thereby enhancing player immersion by balancing predictable game structures with realistic, flexible NPC responses.
The proposed methodology leverages FSMs to manage structured NPC behaviors and state transitions, such as client satisfaction levels based on player actions like response time and solution quality. Concurrently, ABMs are intended to provide dynamic, adaptive reactions to environmental changes and player input, contributing to a more realistic simulation. A preliminary FSM model was developed using JFLAP, successfully demonstrating client mood and satisfaction states. The scope of this work is primarily theoretical and conceptual, with an initial practical application in a software development educational context.
While the study successfully demonstrated the potential of this integrated approach to improve player immersion, it acknowledges significant limitations. The complexity of integrating FSMs and ABMs, the current lack of advanced tools beyond JFLAP for such integration, and the absence of a dedicated development environment pose challenges. Future research is directed towards developing robust frameworks for efficient integration, expanding the complexity of the models, and incorporating machine learning with real-world data to further enhance agent adaptability and accuracy in serious game environments.
European Conference on Games Based Learning (ECGBL) · 2026